The dynamic design of sports vehicles is complicated by the high number of degrees of free- dom (suspension settings, electronic controls, power, aerodynamics, tyres) and their strong interdependencies. In the early stages, traditional methods, which use complex models and track testing, are costly, difficult to transfer, and often struggle to clarify design priorities. This thesis addresses the problem by proposing a simple and interpretable method for linking architectural choices and lap time, in order to guide early decisions with objective and gen- eralisable metrics. The approach breaks down the lap into five phases, namely braking, entry, cornering, exit, and traction, and defines a synthetic KPI for each (average acceleration for “pure” phases and envelope area on the G-G map for “combined” phases) that captures the dynamic effec- tiveness of the phase itself. To explore the design space in a broad but controlled manner, the simulations are conducted on parametric ovals (with variable radius and straights), which replicate the essential components of a circuit while avoiding the complexity of real tracks. Sensitivity analysis is performed on four high-impact parameters, such as tyre size, weight distribution, torque allocation, and elastic set-up, explaining how they act individually and in combination on performance and stability. Data-driven tools (correlations, relative impact, feature selection, and linear regression) are applied to the obtained dataset to extract interpretable importance metrics of the KPIs and therefore of the driving phases. The results indicate that cornering is the main contributor to performance, especially at low speeds, followed by combined phases, while straight-line braking and traction show marginal importance overall and can be excluded. The multi-oval extension and circuit-to-oval mapping demonstrate the adaptability of the method to different scenarios and provide practical guidelines for setting configurations that maximise lateral capacity without penalising entry and exit, enabling quick and informed decisions in the early stages of development.
La progettazione dinamica di veicoli sportivi è complicata dall’alto numero di gradi di lib- ertà (tarature sospensive, controlli elettronici, potenza, aerodinamica, pneumatici) e dalle loro forti interdipendenze. Nelle fasi iniziali, i metodi tradizionali, che fanno uso di modelli com- plessi e testing su pista, sono costosi, poco trasferibili e spesso faticano a chiarire le priorità progettuali. Questa tesi affronta il problema proponendo un metodo semplice e interpretabile per collegare scelte architetturali e tempo sul giro, così da guidare decisioni precoci con met- riche oggettive e generalizzabili. L’approccio scompone il giro in cinque fasi, ovvero frenata, ingresso, percorrenza, uscita e trazione, e definisce per ciascuna un KPI sintetico (media delle accelerazioni per le fasi “pure” e area dell’inviluppo sulla mappa G–G per le fasi ”combinate”) che cattura l’efficacia dinamica della fase stessa. Per esplorare in modo ampio ma controllato lo spazio di progetto, le simulazioni sono condotte su ovali parametrici (raggio e rettilinei variabili), che replicano le componenti essenziali di un circuito evitando la complessità dei tracciati reali. La sensi- tività è svolta su quattro parametri ad alto impatto, quali taglia pneumatici, distribuzione dei pesi, ripartizione di coppia e assetto elastico, esplicitando come agiscono singolarmente e in combinazione su performance e stabilità. Sul dataset ottenuto si applicano strumenti data-driven (correlazioni, impatto relativo, feature selection e regressione lineare) per ottenere metriche di importanza interpretabili dei KPI e quindi delle fasi di guida. I risultati indicano che la percorrenza è il principale artefice della prestazione, specie a raggi piccoli, seguita dalle fasi in combinato, mentre le componenti di frenata e trazione a ruote dritte mostrano importanza marginale nel complesso e possono es- sere escluse. L’estensione multi-ovale e il mapping da circuito ad ovali mostrano l’adattabilità del metodo a scenari diversi e forniscono linee guida pratiche per impostare configurazioni che massimiz- zano la capacità laterale senza penalizzare ingresso e uscita, consentendo decisioni rapide e motivate nelle prime fasi di sviluppo.
Quantifying KPI Importance: A Data-Driven Approach for Sport-Oriented Four-Wheel Vehicles
MENGO, EDOARDO
2024/2025
Abstract
The dynamic design of sports vehicles is complicated by the high number of degrees of free- dom (suspension settings, electronic controls, power, aerodynamics, tyres) and their strong interdependencies. In the early stages, traditional methods, which use complex models and track testing, are costly, difficult to transfer, and often struggle to clarify design priorities. This thesis addresses the problem by proposing a simple and interpretable method for linking architectural choices and lap time, in order to guide early decisions with objective and gen- eralisable metrics. The approach breaks down the lap into five phases, namely braking, entry, cornering, exit, and traction, and defines a synthetic KPI for each (average acceleration for “pure” phases and envelope area on the G-G map for “combined” phases) that captures the dynamic effec- tiveness of the phase itself. To explore the design space in a broad but controlled manner, the simulations are conducted on parametric ovals (with variable radius and straights), which replicate the essential components of a circuit while avoiding the complexity of real tracks. Sensitivity analysis is performed on four high-impact parameters, such as tyre size, weight distribution, torque allocation, and elastic set-up, explaining how they act individually and in combination on performance and stability. Data-driven tools (correlations, relative impact, feature selection, and linear regression) are applied to the obtained dataset to extract interpretable importance metrics of the KPIs and therefore of the driving phases. The results indicate that cornering is the main contributor to performance, especially at low speeds, followed by combined phases, while straight-line braking and traction show marginal importance overall and can be excluded. The multi-oval extension and circuit-to-oval mapping demonstrate the adaptability of the method to different scenarios and provide practical guidelines for setting configurations that maximise lateral capacity without penalising entry and exit, enabling quick and informed decisions in the early stages of development.| File | Dimensione | Formato | |
|---|---|---|---|
|
Mengo_Edoardo.pdf
Accesso riservato
Dimensione
6.33 MB
Formato
Adobe PDF
|
6.33 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/94112